feat(build-all): add trim_pkgcache_metadata helper to bound pkgcache _metadata

local/r-minor-helpers.R

@@ -31,3 +31,42 @@ parse_build_args <- function(args) {
     ncpus = as.integer(pos[3L])
   )
 }
+
+# Bound the {pkgcache} metadata dir, which otherwise grows without limit: the
+# "patched" repo mints a new content hash on every PACKAGES change, so each build
+# writes a fresh ~70 MB _metadata/pkgs-<hash>.rds (+ patched-<hash>/) that is
+# never reused. Keep the `keep` newest entries by mtime; only remove entries
+# older than `min_age_secs`, so a concurrent split-job's in-flight files are
+# never deleted (each build uses a unique hash, so aged entries are
+# unreferenced). Stable repo dirs (CRAN-*, BioC*, INLA-*) and pkg/ downloads are
+# not matched and thus preserved. Returns the number of entries removed.
+trim_pkgcache_metadata <- function(cache_dir = Sys.getenv("R_PKG_CACHE_DIR"),
+                                   keep = 20L,
+                                   min_age_secs = 600) {
+  meta <- file.path(cache_dir, "R", "pkgcache", "_metadata")
+  if (!nzchar(cache_dir) || !dir.exists(meta)) {
+    return(0L)
+  }
+  entries <- c(
+    Sys.glob(file.path(meta, "patched-*")),
+    Sys.glob(file.path(meta, "pkgs-*.rds"))
+  )
+  if (length(entries) == 0L) {
+    return(0L)
+  }
+  info <- file.info(entries)
+  order_new_first <- order(info$mtime, decreasing = TRUE)
+  ranked <- entries[order_new_first]
+  ranked_mtime <- info$mtime[order_new_first]
+  if (length(ranked) <= keep) {
+    return(0L)
+  }
+  candidates <- ranked[(keep + 1L):length(ranked)]
+  candidate_age <- as.numeric(Sys.time()) - as.numeric(ranked_mtime[(keep + 1L):length(ranked)])
+  removable <- candidates[candidate_age >= min_age_secs]
+  if (length(removable) == 0L) {
+    return(0L)
+  }
+  unlink(removable, recursive = TRUE, force = TRUE)
+  length(removable)
+}

local/tests/test-trim-pkgcache.R

@@ -0,0 +1,68 @@
+source(file.path("..", "r-minor-helpers.R"))
+
+# Build a fake _metadata dir under a temp R_PKG_CACHE_DIR. Each entry's mtime is
+# set to `age_secs` in the past so we can exercise the age gate deterministically.
+make_meta <- function(patched = 0L, pkgs = 0L, keep_repos = TRUE, age_secs = 3600) {
+  root <- tempfile("pkgcache-")
+  meta <- file.path(root, "R", "pkgcache", "_metadata")
+  dir.create(meta, recursive = TRUE)
+  old <- Sys.time() - age_secs
+  mk_dir <- function(p) { dir.create(p); Sys.setFileTime(p, old); p }
+  mk_file <- function(p) { writeLines("x", p); Sys.setFileTime(p, old); p }
+  for (i in seq_len(patched)) mk_dir(file.path(meta, sprintf("patched-%03d", i)))
+  for (i in seq_len(pkgs)) mk_file(file.path(meta, sprintf("pkgs-%03d.rds", i)))
+  if (keep_repos) {
+    mk_dir(file.path(meta, "CRAN-075c426938"))
+    mk_dir(file.path(meta, "BioCsoft-1ac964ed6c"))
+    mk_file(file.path(meta, "bioc-sysreqs.dcf.gz"))
+    mk_dir(file.path(root, "R", "pkgcache", "pkg")) # downloads, must survive
+  }
+  root
+}
+
+n_churn <- function(root) {
+  meta <- file.path(root, "R", "pkgcache", "_metadata")
+  length(Sys.glob(file.path(meta, "patched-*"))) +
+    length(Sys.glob(file.path(meta, "pkgs-*.rds")))
+}
+
+test_that("empty cache_dir is a no-op", {
+  expect_identical(trim_pkgcache_metadata("", keep = 5L, min_age_secs = 0), 0L)
+})
+
+test_that("missing _metadata dir is a no-op", {
+  expect_identical(
+    trim_pkgcache_metadata(tempfile("absent-"), keep = 5L, min_age_secs = 0),
+    0L
+  )
+})
+
+test_that("fewer than keep entries removes nothing", {
+  root <- make_meta(patched = 2L, pkgs = 2L)
+  expect_identical(trim_pkgcache_metadata(root, keep = 20L, min_age_secs = 0), 0L)
+  expect_identical(n_churn(root), 4L)
+})
+
+test_that("trims down to keep newest, leaving churn == keep", {
+  root <- make_meta(patched = 30L, pkgs = 30L) # 60 churn entries, all old
+  removed <- trim_pkgcache_metadata(root, keep = 20L, min_age_secs = 0)
+  expect_identical(removed, 40L)
+  expect_identical(n_churn(root), 20L)
+})
+
+test_that("entries younger than min_age_secs are protected", {
+  root <- make_meta(patched = 30L, pkgs = 0L, keep_repos = FALSE, age_secs = 60)
+  # keep=5 would drop 25, but all are 60s old < 600s gate -> nothing removed
+  expect_identical(trim_pkgcache_metadata(root, keep = 5L, min_age_secs = 600), 0L)
+  expect_identical(n_churn(root), 30L)
+})
+
+test_that("stable repo dirs and pkg downloads are never touched", {
+  root <- make_meta(patched = 30L, pkgs = 30L)
+  trim_pkgcache_metadata(root, keep = 0L, min_age_secs = 0)
+  meta <- file.path(root, "R", "pkgcache", "_metadata")
+  expect_true(dir.exists(file.path(meta, "CRAN-075c426938")))
+  expect_true(dir.exists(file.path(meta, "BioCsoft-1ac964ed6c")))
+  expect_true(file.exists(file.path(meta, "bioc-sysreqs.dcf.gz")))
+  expect_true(dir.exists(file.path(root, "R", "pkgcache", "pkg")))
+})